环境科学与技术最新文献

Black carbon (BC) variability in coastal environmentsreflectsthe combined effects of emission sources, regional transport, andmesoscale circulation. This study selected the typical coastal cityof Qingdao as the study area, collected BC online observation datafor one year in 2021, and integrated it with the Aethalometer model,trajectory analysis, and Random Forest (RF) machine learning frameworkto examine these mechanisms. The annual mean BC concentration was1.78 ± 1.29 μg m^–3, with fossil fuelcombustion contributing 88.2% of the total mass, while the biomassburning contribution increased to 19% in winter. Source-resolved attributionshowed distinct meteorological responses. Fossil-fuel black carbon(BC_ff) was more strongly associated with wind-controlledventilation and transport, whereas biomass burning BC (BC_bb) responded more strongly to low-temperature conditions. Under sea–landbreeze (SLB) conditions, BC enhancement was most evident at wind speedsof 2.0 to 3.5 m s^–1. Diurnal persistence, directionalshifts in elevated fossil fuel BC, and interactions between wind speedand wind direction further supported return transport and reaccumulationwithin coastal circulation. These findings provide process-based insightinto source-dependent BC variability and SLB-related reaccumulationin coastal environments.

Chemical exposure causes adverse neurological outcomes,highlightingthe need for high-throughput screening of chemical neurotoxicity.However, existing computational methods rely on single-modality inputsand lack complementary molecular information to sufficiently enhancethe accuracy and efficiency of neurotoxicity risk assessment. Here,we present NeuroToxPredictor, a scalable and interpretable multimodalframework built on a curated chemical neurotoxicity dataset for accurateand efficient chemical neurotoxicity prediction. It integrates graphattention networks and molecular fingerprints with ChemBERTa, a pretrainedchemical language model that provides contextual encoding of SMILES.A gated fusion module adaptively weights these complementary representationsaccording to their predictive relevance to enhance modality-specificweight allocation through a gating fusion mechanism. NeuroToxPredictorachieved competitive performance across five independent trainingseeds (AUC = 0.919 ± 0.004; accuracy = 0.845 ± 0.014; F₁-score = 0.861 ± 0.011; recall = 0.884 ± 0.013;precision = 0.839 ± 0.024; MCC = 0.688 ± 0.028), outperformingboth single-modality and reduced-modality baselines. It demonstratesstrong generalizability with an AUC of 0.861, as externally validatedon an independent set of 279 compounds. NeuroToxPredictor is deployedon a publicly platform (https://www.ai4environ.cn/NeuroToxPredictor), enabling an end-to-end intelligent workflow from chemical retrieval,neurotoxicity prediction, applicability-domain assessment, and largelanguage model-assisted result interpretation.

Tire rubber-derived chemicals (TRCs) are known to behighly toxicto fish, but understanding of their environmental levels and fateis still limited. This study presents a new method for the analysisof 18 TRCs, including phenylenediamines (TRC-PDs), their quinone derivatives(TRC-Qs), and rubber vulcanizing and curing-related chemicals (TRC-VCRs)in fish bile and plasma samples. We applied this method in bile andplasma samples from English sole collected in urban, near-urban, andnonurban sites in Puget Sound, WA, USA. TRCs were detected in 100%of bile and 96% of plasma samples from English sole collected fromall nine sites in Puget Sound. Mean TRC levels in bile samples persite ranged between 0.8 and 18 ng/mL, while plasma levels ranged between0.22 and 0.63 ng/mL, with total summed reaching up to 31 ng/mL inbile and 1.3 ng/mL in plasma. TRC-PDs were the most abundant groupof TRC in both bile and plasma. Higher levels of TRCs in bile suggeststhis matrix to be more valuable for environmental monitoring, providingmore sensitive detection of TRC during low exposures. To the bestof our knowledge, this is the first data set reporting environmentallevels of 18 TRCs in the bile and plasma of field-caught fish.

Schrieveret al. (2025) argue that environmental risk assessmentof pesticides in the European Union is sufficiently protective andthat regulatory thresholds are rarely exceeded. Here, we reexaminethese claims based on existing evidence from monitoring, systematicreviews, and different types of field studies. The available evidenceindicates that measured pesticide concentrations frequently exceedpredicted concentrations and regulatory thresholds and are associatedwith adverse ecological impacts. The mechanisms driving this resultinclude the joint action of pesticides applied in temporal and spatialproximity, varied environmental conditions, interactive effects ofpesticides with additional stressors, and indirect ecological effectsthat propagate within biological communities. These neglected intricaciesexplain why current single-substance prospective assessments tendto underpredict real-world pesticide exposure and impact in fieldsettings. The corollary is that the current regulatory framework inEurope proves to be insufficient to protect biodiversity and ecosystems.Priority regulatory reforms include strengthening postregistrationmonitoring, refining predictive exposure models, and explicitly consideringlandscape context, indirect and mixture effects, and interactionswith nonchemical stressors.

Ball-milled sulfidated microscalezerovalent iron (S-mZVI^bm) is highly effective in degradingtrichloroethylene (TCE) in anoxicwater. This study investigated its performance and soil-mediated regulationin a soil-water mixed system (SWMS) using two natural soils. The resultsshow that soil addition markedly enhanced the reactivity of S-mZVI^bm. Compared with the aqueous system, the degradation rateconstant of TCE (k_TCE) exhibited a 6.0–8.4-foldincrease at a soil-to-water ratio (S/W) of 0.5 to 2.0, while H₂ production concurrently increased 210%–450%. Thisactivation was accompanied by a decrease in apparent electron efficiencyfrom 74.3% in water to 15.9%–27.6% in SWMS. S-mZVI^bm retained higher reactivity in SWMS than in water across pH 6–10,with a maximum k_TCE of 0.26 h^–1 at pH 8.0. SEM, XPS, and component-specific experiments indicatedthat soil-mediated surface restructuring and Fe(II) redistribution-induceddepassivation increased the accessibility of Fe⁰-rich domains,whereas organic matter and Ca/Mg-bearing phases imposed inhibitoryeffects. Sulfidation suppressed H₂ evolution and reducedsoil-specific variability under multicontaminant conditions involvingTCE, carbon tetrachloride, nitrobenzene, and perchloroethylene. Theseresults identify soil as an active regulator of Fe⁰ accessibilityand electron allocation rather than a passive reaction medium, providinga mechanistic basis for site-specific S-mZVI^bm remediation.

Vehicle-to-grid (V2G) is an important flexibility managementtechnologyfor lower-carbon transport and the electricity grid. However, itseffectiveness relies on the temporal alignment between the price incentivesand grid dynamics. To overcome the challenge of coordinating electricitypricing with V2G deployment, we integrated a Bayesian optimizationframework into a unit commitment model of the 2030 Jing–Jin–Tangpower grid. Results demonstrate that current static time-of-use tariffsincentivize EV fleets to discharge during midday solar peak hours.This behavior displaces zero-marginal-cost renewable generation andforces pumped storage hydropower units to absorb excess fleet discharge.Optimizing tariffs could reduce average net load, operating costs,and CO₂ emissions. In summary, the synergistic interactionbetween V2G and dynamic pricing shows a dual effect: it reduces costsfor both the grid and EV owners but, conversely, can increase CO₂ emissions due to excessive EV participation. Our resultsdemonstrate that a shift toward dynamic, carbon-aware pricing mechanismsis essential to ensure that widespread V2G deployment effectivelysupports decarbonization goals.

Undocumented orphaned oil and gas wells can pose risksthroughunintended releases of substances, yet they remain difficult to locatebecause records are incomplete and surface indicators are often absent.Unmanned aerial vehicle (UAV) magnetometry is increasingly used toidentify steel-cased wells, but surveys commonly stop at target detectionand provide limited guidance on how sensor height affects anomalystrength, burial-depth estimates, or confidence in non-detection.Here, we develop a physics-informed, calibrated empirical model betweenpeak magnetic-anomaly amplitude and distance from the magnetometerto the top of steel casing. Using anomaly amplitude measured at knownsensor height, the model provides three practical outputs: estimatedcasing-top burial depth where casing is not visible, maximum sensorheight at which casing at a specified depth remains detectable, andcorrection of variable-height anomalies to a common reference height.We calibrate the relationship using multi-altitude UAV surveys atWildcat Canyon Regional Park, California, and Osage Ranch, Oklahoma,and apply it at Hillman State Park, Pennsylvania, identifying oneundocumented-well candidate supported by ground magnetometry and estimatingburial depths for three documented wells consistent with independentdepth constraints.